Назад
Company hidden
12 дней назад

Master's Thesis: Robot Learning from Operational Data (Robotics)

Формат работы
onsite
Тип работы
project
Грейд
trainee
Английский
b2
Страна
Sweden
Вакансия из списка Hirify.GlobalВакансия из Hirify Global, списка международных tech-компаний
Для мэтча и отклика нужен Plus

Мэтч & Сопровод

Для мэтча с этой вакансией нужен Plus

Описание вакансии

Текст:
/
TL;DR
Master's Thesis: Robot Learning from Operational Data (Robotics): Investigating whether operational rollout data can improve Vision-Language-Action policies for real assembly tasks with an accent on dataset characterization, fine-tuning experiments, and outcome-label reliability. Focus on comparing operational and teleoperation data, testing filtering and weighting strategies, and evaluating policy quality changes using existing robot datasets.

Location: Stockholm, Sweden; onsite

Company

hirify.global builds Factory OS, a software layer for humanoid robots such as the Unitree G1-D to perform high-mix, low-volume assembly work.

What you will do

  • Compare operational rollout data with curated teleoperation datasets by volume, success and failure balance, and recurring failure modes.
  • Test whether fine-tuning Vision-Language-Action policies on operational data changes policy quality.
  • Evaluate filtering and weighting based on operator-judged outcome labels.
  • Investigate additional data-centric questions, including the reliability and consistency of outcome labels.
  • Characterize the value of operational data for future robot training and provide an evidence-based recommendation.

Requirements

  • Master's student in machine learning or robot learning.
  • Comfortable with empirical experimentation and data analysis.
  • Interest in investigating robot learning for real assembly tasks.

Culture & Benefits

  • Onsite thesis hosted by hirify.global AB in Stockholm within the Robotics department.
  • Technical supervision from a company supervisor.
  • Access to existing operational and curated datasets.
  • Access to the training stack for fine-tuning experiments.
  • Access to a real Unitree G1-D if additional rollouts are needed.

Hiring process

  • Submit a CV and a GitHub, LinkedIn, or personal website link.
  • Explain why the thesis interests you and share an example of work you are proud of.

Будьте осторожны

Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →